Our studies utilize a CHO cell line (AS52) that carries a single functional copy of the bacterial gpt gene stably integrated into the CHO genome. Mutations at the gpt locus can be recovered for molecular studies as 6-thioguanine resistant (TGr) colonies. Independent mutants are characterized resulting in the generation of a mutational spectrum. Point mutational spectra are a well defined collection of mutations that can provide a basis for proposing hypotheses regarding mutational mechanisms. Such spectra are most easily generated using the polymerase chain reaction (PCR) followed by DNA sequence analysis. We have applied techniques that allow a more rapid generation of point mutational spectra. First, independent mutants are isolated in situ followed by PCR amplification of the mutant gpt gene. Then, the site of the mutation within the PCR fragment is localized prior to sequence analysis to minimize the amount of DNA sequencing required. For the in situ selection protocol, cells are plated in expression medium containing low melting point (LMP) agarose immediately following treatment. Cells are incubated to allow phenotypic expression before being overlaid with 6TG containing selection medium. Using this protocol, all TGr colonies arising in the agarose are independent mutants, thus simplifying the isolation of mutants for molecular analyses. The site of the mutation in PCR fragments is identified prior to DNA sequence analysis by chemical modification and cleavage at the site o a mismatch in a (32)P-end labeled wild-type/mutant-DNA heteroduplex. The cleaved (32)P-end labeled fragment can then be sized on a denaturing-acrylamide gel allowing precise localization of the site of the point mutation. The utility of these protocols has been demonstrated in the analysis of mutations induced by UV, Actinomycin D, Formaldehyde, and Mitomycin C. The in situ selection is generally applicable to other mammalian mutational assay systems and the mismatch analysis not only facilitates rapid sequence determination of the site of the mutation but may be especially useful in identifying the sites of mutations in much larger structural genes (e.g., p53) where routine sequencing not the entire gene is not practical.